Cold springs can only indicate the mean atmospheric temperature p 221 when they are unmixed with the waters rising from great depths, or descending from considerable mountain elevations, and when they have passed through a long course at a depth from the surface of the earth which is equal in our latitudes to 40 or 60 feet, and according to Boussingault, to about one foot in the equinoctial regions,* these being the depths at which the invariability of the temperature begins in the temperate and torrid zones, that is to say, the depths at which horary, diurnal, and monthly changes of heat in the atmosphere cease to be perceived. [footnote] * The profound investigations of Boussingault fully convince me, that in the tropics, the temperature of the ground, at a very slight depth, exactly corresponds with the mean temperature of the air. The following instances are sufficient to illustrate this fact: ________________________________________________________ Stations Temperature at Mean Height, in within 1 French foot Temperature English Tropic [1.006 of the of the feet, above Zones. English foot] air. the level below the of the sea. earth's surface. ________________________________________________________ Guayaquil 78.8 78.1 0 Anserma Nuevo 74.6 74.8 3444 Zupia 70.7 70.7 4018 Popayan 64.7 65.6 5929 Quito 59.9 59.9 9559 ________________________________________________________ The doubts about the temperature of the earth within the tropics, of which I am probably, in some degree, the cause, by my observations on the Cave of Caripe (Cueva del Guacharo), 'Rel. Hist.', t. iii., p. 191-196), are resolved by the consideration that I compared the presumed mean temperature of the air of the convent of Caripe, 65.3 degrees, not with the temperature of the air of the cave, 65.6 degrees, but with the temperature of the subterranean stream, 62.3degrees, although I observed ('Rel. Hist.', t. iii., p. 146 and 195) that mountain water from a great height might probably be mixed with the water of the cave.